I’m not too worried by AI. I don’t make music to earn a living, I make music because I must. Getting paid for a track or gig here or there is really a bonus. Most artists can’t live from their art, it’s how it is.
AI for now is constrained by the data it gets fed, humans don’t have such constraints. Creativity is hard if not impossible to program. Music created by AI might be ok for many people as it can learn from millions of hit records and create new pieces based on that data. But it will be limited.
On a side note, I want to feed an AI only Captain Beefheart songs and see what it comes up with. Are there any websites where we can play with AI to generate music yet?
The channel has ai generated death and trash metal aswell, some irritating stuff : hal 9000 sings carcass songs, spooky as hell … but you see ai is making progress.
He lost two children tragically which makes me feel both sad and offended by those lyrics made by AI and by the person who thought it was a good idea to do such a thing and post them online.
In research, we’re starting to see a limited impact of algorithms on some working practices.
For example, qualitative research relies on transcription of audio recordings of interviews or focus groups. In the past, we might have paid £150 per hour of recording (approx) to a professional transcriber to do the work, and hence 100 interviews came with a cost of £15k. The cost of transcribing was a major bottleneck on the range of interviews that could be done.
Now, transcription costs have dropped to £60 per hour approx, as the work of a transcriber will involve running an algorithm to get a rough transcript, which the transcriber then corrects. The result is that we can do more interviews for the same transcription cost, and hence generate a greater depth of knowledge on a topic. The transcriber is also liberated from the most repetitive elements of their work. However, I wouldn’t be surprised if a lot of transcribers have lost their work through this change.
In music, I suspect that we will see even more of a squeeze on production incomes. There will be lots of lower end producers who will be squeezed out, such as those doing jingles for radio shows. This is just accentuating a trend in which producer income has been squeezed by the mass availability of a back catalogue of electronic music.
I do wonder whether that trend is already creating change in the electronic music industry. Its been fascinating seeing people like Goldie creating live shows with full bands. I need to see one of those. Perhaps change like that will be how some people retain an income.
Of course, the small band playing a gig in a rented hall in a warehouse somewhere will be unaffected, and that kind of raw music scene will always appeal to some (like me).
Nothing more than disposable music for disposable people. Much like when having to deal with the soulless rubbish blaring at the gym, I’ll put on a pair of noise cancelling headphones and tune out.
It seems inevitable that we’ll have some kind of prompt based music AI that will work especially well for electronic music. And this music will probably suffice in some commerical applications. I think whether hobbyists or professional musicians, we should have the utmost sympathy for anyone whose income or livelihood is impacted by this. As we know, technolical innovation isn’t leading to some utopia where machines do the boring stuff and we all spend most of the time engaging in leisure activities, learning, and creating art.
Whether AI can make hit songs is another interesting question because, making a broad assumption here, most people like music they recognise and that enables some kind of social bonding with others: “I love this song, Dave. Haven’t heard it in years…” Wedding bands, for example, don’t generally play obscure songs or original material.
music i listen, can’t be made by computer. AI is man made.
p.s
all that recent talk about ai reminds me on my kid playing with dolls, lending them personality, and then pretending the toy has life of its own.
Man the vocals are some uncanny valley stuff. It is literally the auditory equivalent of those ai images where everything looks real but you can’t name what anything is.
